Guide
Pentest retesting: how to verify fixes properly
A retest is where a penetration test pays off: it proves the vulnerabilities are actually gone. Here's how to scope it, run it, price it and document it so the result holds up with clients and auditors.
Updated 28 October 2026 · By the Closeout team
What is a pentest retest?
A retest (also called a remediation test or verification test) is a follow-up in which the tester checks whether the vulnerabilities reported in a penetration test have been fixed. It is not a new pentest: it targets the reported findings, starts from the original evidence, and ends with each finding marked as fixed, partially fixed or still vulnerable.
Without a retest, the report only says what was wrong on the day of testing. With one, the client can show that the problems were found and resolved, which is what auditors and customers actually want to know.
What a retest should cover
- Every finding the client says is fixed, replaying the original proof of concept.
- Variants of the same issue. A fix for one parameter often misses the next one: an XSS filtered on
/searchmay still work on/profile. - The fix itself. Did it introduce something new, such as an error page that now leaks a stack trace?
- Findings accepted as a risk are not retested, but they are recorded with who accepted them and when.
What it should not cover: new features, new assets or a fresh hunt for vulnerabilities. If the scope changed, that is a new test with its own authorization.
When to retest
As soon as the client says a fix is deployed, while the context is fresh on both sides. Retests batched weeks later cost more: environments change, test accounts expire and the tester has to rebuild context from the report. Contracts commonly include one retest within 30 to 90 days of the report; our guide to remediation timelines covers how long each severity usually gets.
How to run a retest, step by step
- Confirm the fix is deployed to the environment that was in scope, and note the version or commit.
- Check the authorization still covers today. Retesting outside the agreed window is still unauthorized testing.
- Replay the original proof of concept exactly as reported.
- Try the obvious bypasses and variants: other parameters, encodings, HTTP methods, roles.
- Capture evidence of the new behavior: the request, the response and a timestamp.
- Set the new status, and have a second person review critical and high findings before they are closed.
- Issue a new version of the report (v2) rather than overwriting the original, and update the attestation letter.
What counts as “fixed”
| Status | Meaning | Example |
|---|---|---|
| Fixed | The root cause is addressed; the PoC and its variants fail | Parameterized queries replace string concatenation |
| Mitigated | Exploitation is blocked, but the cause remains | A WAF rule blocks the payload; the code is still injectable |
| Partially fixed | Some instances are fixed, others are not | XSS fixed on three of five parameters |
| Not fixed | The PoC still works | Same response as in the original test |
| Risk accepted | The client formally chose not to fix it | A legacy system scheduled for decommissioning |
Be careful with “mitigated”. A WAF rule can be a valid compensating control, but the report and the letter should say so instead of calling the finding fixed.
How to price retests
- Included retest: one retest within a set number of days is part of the original price. The most common model for boutique firms; many budget it at a small share of the original effort.
- Time and materials: billed by the hour or day on top. Fair for large reports with many findings.
- Per finding: predictable for the client, easy to quote.
- Unlimited within a window: a strong differentiator, but only viable when each retest costs you little, which is where automation comes in.
How to document the result
Update each finding with the retest date, the tester, the evidence and the new status. Re-issue the report as a new version, keep the original, and give the client a one-page summary they can share. Our free attestation letter generator writes that letter for you, including the retest results by severity.
Automating retests safely
Many web and API findings can be retested mechanically: the proof of concept is an HTTP request, and the vulnerable or fixed signal is in the response. Automating those saves hours per engagement and lets you retest the moment a fix ships, instead of when the calendar allows.
Three rules keep it safe: requests only ever go to hosts the client authorized, inside the testing window; they are rate limited; and a person confirms every result before a finding is closed.
Frequently asked questions
How long after a pentest should the retest happen?
Usually within 30 to 90 days of the report, and ideally as soon as each fix is deployed, while the context is fresh.
Is a retest a new penetration test?
No. A retest only checks whether the reported findings were fixed. New features, new assets or a broader search for vulnerabilities are a new test.
Does a retest need a new authorization?
The written authorization must cover the dates of the retest. If the original testing window has ended, extend it in writing before retesting.
Who should perform the retest?
Ideally the same tester or firm, since they already have the context and the original evidence. An independent retest by another firm is effectively a new engagement.
What if a fix introduces a new vulnerability?
Report it as a new finding, with its own severity. Whether it is fully tested under the current engagement is something to agree with the client.